Interface Driven Energy Filtering of Thermoelectric Power in SparkPlasma Sintered Bi2Te2.7Se0.3 Nanoplatelet CompositesAjay Soni,† Yiqiang Shen,‡ Ming Yin,‡ Yanyuan Zhao,† Ligen Yu,§ Xiao Hu,‡ Zhili Dong,‡

Khiam Aik Khor,§ Mildred S. Dresselhaus,∥ and Qihua Xiong*,†,⊥

†Division of Physics and Applied Physics, School of Physical and Mathematical Sciences, Nanyang Technological University,Singapore 637371, Singapore‡School of Material Science and Engineering, Nanyang Technological University, Singapore 639798, Singapore§Division of Manufacturing Engineering, School of Mechanical and Aerospace Engineering, Nanyang Technological University,Singapore 639798, Singapore∥Department of Physics and Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology,Cambridge, Massachusetts 02139, United States⊥Division of Microelectronics, School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639798,Singapore

ABSTRACT: Control of competing parameters such as thermo-electric (TE) power and electrical and thermal conductivities isessential for the high performance of thermoelectric materials. Bulk-nanocomposite materials have shown a promising improvement inthe TE performance due to poor thermal conductivity and chargecarrier filtering by interfaces and grain boundaries. Consequently, ithas become pressingly important to understand the formationmechanisms, stability of interfaces and grain boundaries along withsubsequent effects on the physical properties. We report here theeffects of the thermodynamic environment during spark plasmasintering (SPS) on the TE performance of bulk-nanocomposites ofchemically synthesized Bi2Te2.7Se0.3 nanoplatelets. Four pellets ofnanoplatelets powder synthesized in the same batch have beenmade by SPS at different temperatures of 230, 250, 280, and 350°C. The X-ray diffraction, transmission electron microscopy, thermoelectric, and thermal transport measurements illustrate thatthe pellet sintered at 250 °C shows a minimum grain growth and an optimal number of interfaces for efficient TE figure of merit,ZT∼0.55. For the high temperature (350 °C) pelletized nanoplatelet composites, the concurrent rise in electrical and thermalconductivities with a deleterious decrease in thermoelectric power have been observed, which results because of the grain growthand rearrangements of the interfaces and grain boundaries. Cross section electron microscopy investigations indeed showsignificant grain growth. Our study highlights an optimized temperature range for the pelletization of the nanoplatelet compositesfor TE applications. The results provide a subtle understanding of the grain growth mechanism and the filtering of low energyelectrons and phonons with thermoelectric interfaces.

KEYWORDS: Thermoelectric figure of merit, Bi2Te2.7Se0.3 nanoplatelet composites, spark plasma sintering, interfaces, grain boundaries,energy filtering

Recently, the bulk nanocomposite approach has beenshown to be advantageous over their bulk counterparts

to achieve an enhancement in the high thermoelectric (TE)figure of merit, primarily due to poor thermal conductivity orquantum confinement effects although the electrical con-ductivity of those highly disordered nanomaterials is usuallycompromised.1,2 In this context, the designing and engineeringof interfaces and preferential phonon scattering centers innanocomposites have become an important field of research toimprove the performance of future TE materials.3,4 Theefficiency of the TE materials is scaled as a dimensionless TEfigure of merit, ZT = S2σT/(κe + κl), where S is the Seebeck

coefficient, T is the absolute temperature, σ is the electricalconductivity, and κe and κl are, respectively, the electronic andlattice contributions to the thermal conductivity.5 Thus a goodTE material should have a high Seebeck coefficient, a highelectrical conductivity, and a low thermal conductivity.Combining all three physical parameters, doped semiconduc-tors are found to be the best TE materials.5−7 Thoughnanostructure composites are promising candidates for

Received: May 28, 2012Revised: July 14, 2012Published: July 23, 2012

Letter

pubs.acs.org/NanoLett

© 2012 American Chemical Society 4305 dx.doi.org/10.1021/nl302017w | Nano Lett. 2012, 12, 4305−4310

enhanced TE properties,8,9 yet the thermodynamical instabilityof enormous grain boundaries and interfaces limits theirapplication range and reproducibility of performance for TEapplications.6 Therefore, to optimize the processing parametersand to identify the effective range of applicability, it issignificant to study how the thermodynamical environment ofnanocomposite preparations affects their physical properties.The chalcogenides are among the most studied TE materials

and their application spans over the whole temperature range of100−1000 K, with every material representing a typicaltemperature range.5,6,10−12 For instance, Bi2Te3 and relatedcompounds are suitable for low-temperature refrigeration,2

while PbTe, Co4Sb12, and ZnSb based compounds are bestsuited for high temperature power generation.10,13,14 Thetemperature range of TE application is defined by the electronicband structure, crystal structures, and preparation methods,such that the decoupling of all the interdependent physicalparameters (S, σ, and κe + κl) could be possible.14−16 After aresurgence of TE research,13,15 the nanocomposites of Bi2Te3and related compounds have shown potential improvement inZT above 1 for room temperature TE applications with a largenumber of interfaces.2,17 On this note, studies on highperformance bulk nanoplatelet-composites (NPCs) are ofinterest because of facile and scalable synthesis with anisotropicproperties due to grain boundary/interface arrangements.18−22

The interfaces perform an energy-filtering effect that preferen-tially scatters phonons which contribute strongly to the thermalconductivity.3,23,24 In addition, the energy barriers introducedby the interfaces also block the low-energy electrons andtherefore decrease the average heat transported per carrier.3,25

Thus, the introduction of interfaces exhibits significant promiseto suppress thermal conductivity with minimized decreasing ofelectrical conductivity, leading to an improved ZT. The lowenergy charge carrier filtering by interfaces has beendemonstrated in chalcogenide materials composed of thinfilm superlattices,1 nanograins,26 heterostructures,24 and nano-inclusions.27,28 From the comparison of NPCs18,29 withnanoparticles composites,2 it has been understood thatnanoparticles might not be as effective as a filter asnanoplatelets in bulk nanocomposites, because the low energyelectron wave functions might go around the nanoparticles.8

However, for the present case of layered NPs, the chargecarriers and phonons seem to couple in a unique fashion suchthat interfaces can enhance the electronic and thermal transportproperties, anisotropically.3,18 Additionally, for the TE nano-composites, the microstructure, mean-particle-size, particle sizedistribution, materials densification, and grain-boundary prop-erties are very important parameters. Therefore, it is importantto optimize the processing conditions so that a proper choice ofmaterial with higher and more stable ZT could be produced.Previously, we have reported on the enhanced TE properties

of solution grown composites of Bi2Te3−xSex nanoplatelet witha ZT of 0.54 for Bi2Te2.7Se0.3 composites.18,30 In the presentreport, we highlight the effects of the thermodynamicalenvironment, during the preparation of the composites, onthe physical properties of the NP composites. With the TE andthermal transport measurements, a SPS temperature of 250 °Chas been optimized for comparatively stable ZT ∼ 0.55, whichis within the error limit in comparison to the previous report.18

The current study emphasizes the energy filtering due to a largenumber of anisotropic interfaces and grain boundaries in thebulk-NPCs and the effect of the SPS temperature on the typicaltemperature range of their applications, which has not been

addressed extensively in the literature to the best of ourknowledge.Our prior work on the polyol synthesis method has produced

high quality crystalline NPs of Bi2Te3, Bi2Se3, and Se dopedternary Bi2Te3−xSex (0 ≤ x ≤ 3).18,30 We have shown that thecost-effective polyol method can be easily scaled up to a fewgram quantity of the NPs in each growth. More details of thesynthesis and washing of the NPs can be found in the previousliterature.18,30 Four different samples, named as (i) SPS-230C,(ii) SPS-250C, (iii) SPS-280C, and (iv) SPS-350C, have beenpelletized in a graphite die by applying a pressure of 40 MPausing SPS at various temperatures (i) 230 °C, (ii) 250 °C, (iii)280 °C, and (iv) 350 °C, respectively. The SPS chamber waskept under a vacuum of 0.1 mbar after purging with nitrogen toavoid oxidation during pelletization. Disc shape pellets with a12 mm diameter and a thickness of 0.3−0.5 mm were obtained.The mass densities of the pellets are estimated to be greaterthan 87% and found to increase with increasing SPStemperatures; however this is not the case for SPS-230C,where the density is roughly 83% of bulk value. It is believedthat high current during SPS produces sparks within the grainboundaries of NPs and these sparks connect the weak links ofNPs by local heating, without much affecting the morphologiesof the samples. However, for the present NPCs, the SPSoperating at higher temperatures yields grain growth and thussignificant modification to the physical properties of the NPCs.The crystalline structure of the Bi2Te2.7Se0.3 compounds,

similar to the rhombohedral families of Bi2Te3 compounds, hasa layered geometry with each unit cell consisting of threequintuple layers stacked together with van der Waals forces. Aquintuple layer consists of an alternate arrangement of Bi andTe atoms such as Te(1)−Bi−Te(2)−Bi−Te(1), where Teatoms are distinguished with two different chemical environ-ments.31 As reported earlier, Se has complete solubility inBi2Te3 with a preferential occupancy at two different Te sites,with increasing Se concentrations.18 In the case of Bi2Te2.7Se0.3NPs, comparatively more electronegative Se atoms occupyTe(2) sites relative to Bi2Te3 nanocomposites.

18,31 The X-raydiffraction (XRD) patterns of the SPS pellets are shown inFigure 1 (Bruker D8 diffractometer, with a Cu Kα radiation inthe locked-coupled geometry of the X-ray gun and detector).From the comparison of the XRD spectra with the JCPDS datacard no. 50-0954, and the observed prominence of the 006,0010, and 0015 peaks, it is understood that NPs have apreferential orientation in the c-direction. In other words, theseNPs tend to align themselves together during pelletization.18

Figure 1. XRD patterns of the pellets of Bi2Te2.7Se0.3 NPCs sintered atvarious temperatures (as denoted on the right of each trace). Thepresence of 006, 0010, and 0015 diffraction peaks illustrate the typicalorientation of the NPCs.

Nano Letters Letter

dx.doi.org/10.1021/nl302017w | Nano Lett. 2012, 12, 4305−43104306

The lattice parameters estimated from the XRD peaks arefound to be a = 0.437 nm and c = 2.98 nm for SPS-250C,within the error limit for other NPCs.18 The comparativelysharper XRD peaks for SPS-280 and SPS-350 samples illustratethe grain growth at these higher temperatures. It is interestingto note that the 104 peak is appearing gradually for pelletsexcept for sample SPS-230C. For the SPS-350C sample, thepeak corresponding to 104 planes is dominating over the otherdiffraction peaks, which suggests the reorientation of grainsafter crystallization of the interfaces; however, the exact reasonfor such a grain reorientation is not clear at this stage. Tounderstand this effect further, we carried out cross section TEMmeasurements, which will be discussed in the later sections ofthis paper.The morphologies of the NPCs have been carefully examined

by field-emission scanning electron microscopy (FESEM, JEOL7001F). Figure 2 displays representative SEM images of the

cross-section view of the pellets of SPS-230C and SPS-350C(Figure 2a and b). The estimated thickness of the SPS-230CNPs is less than ∼40 nm (Figure 2a), and an apparent graingrowth has been observed (Figure 2b) for SPS-350C with athickness of 150 nm or higher. The top (Figure 2c) and crosssection (figure 2d) views depict the significant orientation ofthe SPS-350C NPCs. It was reported earlier that this typicalorientation affects the physical properties of the NPCssignificantly.18

The temperature-dependent TE transport measurements,using a Quantum Design PPMS instrument, have been carriedout on a 10 mm long bar with a width of 2 mm cut from thecircular disk compacted by SPS. The four-probe method is usedfor electrical resistivity measurements followed by thermo-power and thermal conductivity measurements by heating thesample from one side. In a continuous-mode measurement witha 0.5 K/min ramping rate and a 200 s step pulse applied to theheater, leading to more than 20 min for measuring each datapoint. To avoid thermal losses due to radiation and othersources, the measurements have been performed under a

vacuum of 10−5 mbar. The electrical and TE transport data areshown in Figure 3. The Seebeck coefficient, S (Figure 3a), has a

linear dependence with temperature showing a diffusivetransport for NPCs SPS at low temperatures such as SPS-230C and SPS-250C. The rough estimation of the number ofcharge carriers, using the classical Mott−Jones formula and Hallmeasurements, lies in the range of 1018−1019 cm−3.18 Thetemperature dependence of the electrical resistivity, ρ(T)(Figure 3b), shows a metallic behavior with different phononscattering mechanisms (from the slope dρ/dT) in each pellet.Apparently, the contribution from structural disorder effectscan be estimated qualitatively from the values of the residualresistivities, which decrease for the high temperature SPSNPCs. As expected, the ρ(T) value is reduced for NPCssintered at high temperatures. For instance, the sample SPS-230C has high ρ values at all temperatures, while the samplesSPS-280C (∼36 μΩ·cm) and SPS-350C (∼31 μΩ·cm) have acomparatively better electrical transport in comparison toearlier reports.18,29 This improvement in the electricalconduction for the NPCs is arising from the grain growth orrearrangements of the grain boundaries during SPS at hightemperatures. The ρ and κ are inter-related through theWiedemann−Franz law13,15 and have the similar trends as oneanother for each SPS sample. The contribution from the latticethermal conductivity is also apparent due to the grain growth inthe NPCs. The higher-ρ and smaller-κ values for SPS-230Cresult from the poor connectivity of the NPs to each other andthe low mass density of the NPs.2,13,29 The κ values (in Figure3c) are increasing systematically with the SPS temperatures andapproaching close to the bulk values measured at 300 K of 1.2W/m·K2 for SPS-350C.13,15

The power factor has the highest value for SPS-250C andthen decreases in the sequence of SPS 280C followed by SPS230C and SPS 350C samples. Even though the high

Figure 2. SEM images of the SPS processed Bi2Te2.7Se0.3 NPC pellets.The cross section views of (a) SPS-230C and (b) SPS-350C, which isshowing the partial growth of the NPs; (c) the top view of thescratched piece of SPS-350C pellet and (d) the cross section view ofsmall chipped piece of SPS-350C. Scale bar is 100 nm for a and b and1 μm for c and d.

Figure 3. Temperature dependence of electrical and thermal transportparameters. (a) Seebeck coefficient, (b) electrical resistivity, (c)thermal conductivity, and (d) TE figure of merit and power factor(inset) of SPS pellets of Bi2Te2.7Se0.3 NPCs pelletized at varioustemperatures. All figures share the same color notations, as shown inthe legend of a.

Nano Letters Letter

dx.doi.org/10.1021/nl302017w | Nano Lett. 2012, 12, 4305−43104307

temperature SPS is leading to a better electronic conduction,yet the power factor S2/ρ (inset in Figure 3d) does not improvewith respect to the SPS-250C. The significant reduction of S,for SPS-280C and SPS-350C, appears to be the obvious reasonfor the deprived power factor with the SPS temperatures. Thepower factor for SPS-230C and SPS-350C NPCs is found to benearly equivalent but certainly has different reasons, while thehigh ρ values of SPS-230C is responsible for the poorperformance and the low S and high κ values for SPS-350C.This signifies the need for the present study of complexity inthe nanostructured materials for TE applications. Comparingthe room temperature values, the SPS-250C with a ZT ∼ 0.55has a better TE performance than other NPCs. The roomtemperature ZT value for SPS- 250C is also higher than thereported for the high temperature SPS pelletized bulk NPCs ofBi2Te3 NPs prepared by a solution method.22 The surprisingpoorer ZT for SPS-280C and SPS-350C is because of theconcomitant rise of σ(1/ρ), κ with reduced S values, whereasvery high ρ values, though with high S and low κ, lead to a poorZT for SPS-230C. As a consequence, the unexpected poorvalues of ZT for SPS-280C and SPS-350C demand a furtherinvestigation on the high temperature states and microscopicdetails of the NPCs.During SPS at high temperatures, the evaporation of the

elemental species from the NPCs is suspected. To understandfurther the reason for this speculative evaporation, we haveperformed a thermogravimetry analysis (TGA) of two endsamples, SPS-230C and SPS-350C (Figure 4a). The dataillustrate a maximum mass loss of about 5% at 600 °C. For thepresent NPCs, our concern is the weight loss below 350 °C,which is the highest SPS temperature utilized. The resultsshowed only <1% mass loss below 350 °C. The absence ofimpurity phase peaks in the XRD spectra (Figure 1) suggeststhat there is no significant change in the crystal structure of theNPCs with SPS. Thus less than 1% mass loss is insignificant to

affect the crystal structure of the compounds and might notaccount for the drastic changes in transport properties.Another speculative reason for the poor TE performance is

the rearrangements of grain boundaries and interfaces whichmight give rise to extra entropic contributions from the NPCs.As reported earlier, the interfaces play an important role viafiltering charge carriers in TE nanocomposites.3,4 To under-stand further the effect of high temperature SPS on physicalproperties and to estimate the thermal response of the NPCs,we have carried out differential scanning calorimetry (DSC)measurements in the temperature range of 100−500 °C. ForDSC measurements, a small 2 × 2 mm2 square piece (∼30 mg)was cut from the pellets and was kept in an aluminum panalong with an identical empty one as a reference. Each samplehas gone through a cycle of heating−cooling−heating at therate of 5 °C/s, in the temperature range of 100−500 °C. Anargon atmosphere was maintained inside the chamber to avoidpossible oxidation during heating−cooling cycles. We cau-tiously avoided heating the pellets up to the meltingtemperature of the parent bulk Bi2Te3 (584 °C),13 as thismight cause the evaporation of materials during the experiment.Given that the nanocomposites are ad-mixtures of the grainsand grain boundaries, the heat flow in the NPCs is expected tobe governed by the interfaces and grain boundaries. In DSCmeasurements, the thermodynamic instability of nanostructuresis reflected as an endothermic peak due to the consumption ofthe heat flow in grain growth and as a heat release exothermicpeak for rearrangements or crystallization of the interfaces andgrain boundaries. An irreversible exothermic peak is reportedfor composites of copper and silver nanoparticles where thestored surface energy is released during the first heating cycle ofthe DSC measurements.32,33

Shown in Figure 4b, the SPS-230C has a broad exothermicpeak in the temperature range of 150−500 °C, appearingperhaps due to the crystallization of a large number of

Figure 4. Thermal analysis of the SPS pellets. (a) TGA weight loss versus temperature data of SPS-230C and SPS-350C pellets, showing a less than1% mass loss below 350 °C; (b) DSC curves for the first heating cycle, where a broad thermal response observed for SPS-230C is missing for othersamples; (c) the first cooling and second heating thermal responses of the NPCs (as assigned); however, for clarity the second heating cycle ispresented for SPS-230C only, with a shift in heat flow (other samples also have a similar response). Inset i is a shift in the crystallization (exothermic)peak during the first cooling cycle and the inset ii is showing a melting (endothermic) peak during the second heating cycle. It is speculated from thefirst cooling and second heating that the interfaces are crystallized during the first heating cycle and thus disappear in further heating−cooling cycles.

Nano Letters Letter

dx.doi.org/10.1021/nl302017w | Nano Lett. 2012, 12, 4305−43104308

interfaces between the smaller thickness NPs. As mentionedearlier, the smaller thickness of the NPs in the SPS-230Csample is apparent from the broadness of the XRD peak widthsof the 006, 0010, and 0015 diffraction peaks (Figure 1) andSEM images (Figure 2). With the smaller thickness of the NPs,a higher number of interface and grain boundaries can beexpected in the SPS-230C sample. A very small exothermicpeak at 383 °C is observed for all of the SPS samples. The peakhas been attributed to the presence of Bi2TeO5 withoutconcrete justification.34 Our XRD and small-angle electrondiffraction (SAED) measurements do not support presence ofoxides either. Therefore, the origin of this exothermic peak isstill elusive. For SPS 230C, two endothermic peaks in firstheating at 419 and 449 °C are also observed and attributed tothe eutectic point of Bi−Te phase diagram and melting point ofnonreacted Te, respectively.34 These two peaks, however, arenot seen for the other NPCs, suggesting that the presence of Teis not intrinsic to the other NPCs. We would like to remind thereader here that the NPs are synthesized at the refluxingtemperature of 240 °C, and thus a significant grain growthduring SPS is less likely below 240 °C and is more likely forsamples sintered at temperatures above 240 °C. The plateau-like thermal response up to 370 °C for SPS-250C, SPS-280C,and SPS-350C suggests that the heat flow only rearranges theinterfaces during the heating process. The abrupt rise in heatflow above 430 °C for SPS-250C and SPS-280C may be relatedto the sudden crystallization of several NPs. In the case of theSPS-350C sample, the flat response of the heat flow with adisappearance of the exothermic peak above 400 °C suggeststhat there is only a small amount of involved-contribution froma very much smaller number of interfaces. The crystallization orrearrangements of the interfaces and grain boundaries is anirreversible process, so that when a sample has been treatedwith a heating cycle, the interfacial arrangement is expected tobe locked inside the pellets. To confirm this conjecture, all ofthe NPCs have been given a further slow cooling and reheatingcycle. After the first-heating cycle, all of the NPCs have shown adisappearance of the broad exothermic peak during first-coolingand second-heating (Figure 4c), which is a signature of anirreversible contribution from the stored surface energy or fromthe interface crystallization in the pellets.32,33 The pairing ofexothermic peak in first cooling and endothermic peak insecond heating, for all NPCs, are appearing perhaps because ofthe eutectic temperature of Bi−Te in Bi2Te3 phase diagram.

34 Asystematic variation in the crystallization peak (Figure 4c, inseti) in the thermogram has been observed with respect to theSPS temperature.34 The endothermic peak in the secondheating (Figure 4c, inset ii) does not show variations in peakposition, probably due to the irreversible arrangements of theinterfaces during first heating and cooling. The crystallizationduring the first cooling and the melting in the second heatingwith the absence of a first broad exothermic peak justifies ourargument of rearrangement of interfaces. The crystallization ofthe interfaces at a high temperature supports the transport datawhere both the electrical and the thermal conductivities areincreasing for the high temperature SPS samples. Thus, when asample is sintered at a particular temperature, the arrangementsof the interfaces appear to be locked during the SPS at thattemperature. Since there is no change in the composition, thethermal response of the pellets is arising due to thethermodynamic environment of interface structure lockedinside the NPCs. These interfaces are filtering the low energy

heat and charge carriers in a unique fashion and consequentlyaffecting the physical properties of the NPCs.Now we will discuss the microstructural aspects of the NPCs

to understand the anomalous TE response of the NPCs,investigated by transmission electron microscopy (TEM) andselected area electron diffraction (SAED) measurements (JEOLJEM 2100F) operated at an accelerating voltage of 200 kV (Cs= 0.5 mm, Scherzer defocus = −43 nm). The TEM samples fora cross section view of the pellets were prepared by standarddimpling, followed by a final thinning and polishing by afocused ion beam (Nova Nanolab 600i, FEI). The TEMmicrographs shown in Figure 5 provide an appropriate

explanation and support to the argument of interfacerearrangements and grain growth at high temperature SPS.The grain growth can be perceived from the comparison of thecross section TEM images for samples SPS-230C (Figure 5a)and SPS-350C (Figure 5c). As pointed out earlier, the NPs self-orient along the c-axes and are stacked one over the otherinside the NPCs.9,26 The SAED index pattern, taken on thecross-section of the NPs with the zone axis [21̅1 ̅0] for SPS-250C and [2 ̅110] for SPS-350C, of the rhombohedral crystalstructure shows that the NPs have been grown in the cdirections. The arrows in TEM images are pointing to the caxes [0001]. Thus from the TEM images, the explanation ofgrain growth and interface crystallization, as speculated fromthe TE transport and thermal response of NPCs, is reasonable.It was reported earlier that the NPCs have anisotropic TE

power and the in-plane S is two times higher than the out-of-plane S. This difference is explained on the basis of the differentscattering mechanism in the two directions.18,29 The TEMimages in Figure 6 illustrate the weak link connectivity betweenthe NPs in two different directions of the in-plane and out-of-plane micrographs. It is clear that the top-bottom stacking haslarger area connectivity between the NPs and that thecrystallization of the interfaces is straightforward. On the

Figure 5. Cross section TEM characterizations. (a) TEM image of thesample SPS-250C; (b) a magnified view showing the stacking of NPs.The inset is a SAED pattern of one of the NPs, and the pattern istaken along the zone axis of [21̅1̅0]. (c) TEM image of the sampleSPS-350C; (d) a magnified view showing the grain growth with thecrystallization of the interfaces. The inset is a SAED pattern of one ofthe NPs. The SAED indexing is similar to that in b, however, recordedin a different zone axis of [2 ̅110]. The white arrows correspond to thec-axis [0001].

Nano Letters Letter

dx.doi.org/10.1021/nl302017w | Nano Lett. 2012, 12, 4305−43104309

other hand, the connectivity is poorer in the case of the in-plane head-to-head stacking. In both cases, the interfaces areplaying an important role by forming barrier networks forfiltering the carriers of both the heat and electrical energy;however, the energy filtering from the interfaces is moreeffective with the in-plane transport measurements.In conclusion, we have demonstrated the optimized plasma

sintering conditions and temperature range of the pelletizationfor Bi2Te2.7Se0.3 NPCs for TE applications. The NPCs areoriented along the c-axes and are highly anisotropic. The NPCsintered at 250 °C shows a ZT∼0.55, with the optimal chargecarrier filtering effects originated from the interfaces and grainboundaries. However, the high temperature SPS (350 °C)results in high electrical and thermal conductivities with areduced thermopower and a poor ZT. Thermal transportmeasurements and thermal analysis using TGA and DSCsuggest that the crystallization and rearrangements of theinterfaces are playing an important role in deciding the physicalproperties of the Bi2Te2.7Se0.3 NPCs. The cross section TEMimaging indeed confirms the grain growth and interfacerearrangement. Our results highlight the subtle understandingof engineering interfaces and grain boundaries needed toimprove the TE properties of bulk nanocomposites, particularlyof those layered nanomaterial TE composites.

■ AUTHOR INFORMATIONCorresponding Author*E-mail address: qihua@ntu.edu.sg.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSQ.X. gratefully acknowledges strong support from theSingapore National Research Foundation through a Singapore2009 NRF fellowship grant (NRF-RF2009-06), the SingaporeMinistry of Education via a Tier 2 grant (MOE 2010-T2-2-059), Startup grant support (M58110061), and the NewInitiative Fund (M58110100) from Nanyang TechnologicalUniversity, Singapore. M.S.D. acknowledges support from herDOE/EFRC grant through the S3TEC Research Center atMIT.

■ REFERENCES(1) Venkatasubramanian, R.; Siivola, E.; Colpitts, T.; O’Quinn, B.Nature 2001, 413 (6856), 597−602.(2) Poudel, B.; Hao, Q.; Ma, Y.; Lan, Y.; Minnich, A.; Yu, B.; Yan, X.;Wang, D.; Muto, A.; Vashaee, D.; Chen, X.; Liu, J.; Dresselhaus, M. S.;Chen, G.; Ren, Z. Science 2008, 320 (5876), 634−638.

(3) Medlin, D. L.; Snyder, G. J. Curr. Opin. Colloid Interface Sci. 2009,14 (4), 226−235.(4) Zebarjadi, M.; Esfarjani, K.; Dresselhaus, M. S.; Ren, Z. F.; Chen,G. Energy Environ. Sci. 2012, 5 (1), 5147−5162.(5) Snyder, G. J.; Toberer, E. S. Nat. Mater. 2008, 7 (2), 105−114.(6) Dresselhaus, M. S. Adv. Mater. 2007, 19, 1043−1053.(7) Shakouri, A. Annu. Rev. Mater. Res. 2011, 41 (1), 399−431.(8) Zebarjadi, M.; Esfarjani, K.; Bian, Z.; Shakouri, A. Nano Lett.2010, 11 (1), 225−230.(9) Zebarjadi, M.; Joshi, G.; Zhu, G.; Yu, B.; Minnich, A.; Lan, Y.;Wang, X.; Dresselhaus, M.; Ren, Z.; Chen, G. Nano Lett. 2011, 11 (6),2225−2230.(10) Pei, Y.; Shi, X.; LaLonde, A.; Wang, H.; Chen, L.; Snyder, G. J.Nature 2011, 473 (7345), 66−69.(11) Minnich, A. J.; Dresselhaus, M. S.; Ren, Z. F.; Chen, G. EnergyEnviron. Sci. 2009, 2 (5), 466−479.(12) Liu, H.; Shi, X.; Xu, F.; Zhang, L.; Zhang, W.; Chen, L.; Li, Q.;Uher, C.; Day, T.; Snyder, G. J. Nat. Mater. 2012, 11 (5), 422−425.(13) Nolas, G. S.; Sharp, J.; Goldsmid, H. J. Thermoelectrics: BasicPrinciples and New Materials Developments; Springer: New York, 2001.(14) Kanatzidis, M. G. Chem. Mater. 2009, 22 (3), 648−659.(15) Rowe, D. M. Thermoelectrics handbook: macro to nano; Rowe, D.M., Ed.; CRC/Taylor & Francis: Boca Raton, FL, 2006.(16) Martin, J.; Wang, L.; Chen, L.; Nolas, G. S. Phys. Rev. B 2009, 79(11), 115311.(17) Hicks, L. D.; Dresselhaus, M. S. Phys. Rev. B 1993, 47 (19),12727.(18) Soni, A.; Yanyuan, Z.; Ligen, Y.; Aik, M. K. K.; Dresselhaus, M.S.; Xiong, Q. Nano Lett. 2012, 12 (3), 1203−1209.(19) Mehta, R. J.; Karthik, C.; Singh, B.; Teki, R.; Borca-Tasciuc, T.;Ramanath, G. ACS Nano 2010, 4 (9), 5055−5060.(20) Scheele, M.; Oeschler, N.; Veremchuk, I.; Reinsberg, K.-G.;Kreuziger, A.-M.; Kornowski, A.; Broekaert, J.; Klinke, C.; Weller, H.ACS Nano 2010, 4 (7), 4283−4291.(21) Wenzhong, W.; Poudel, B.; Yang, J.; Wang, D. Z.; Ren, Z. F. J.Am. Chem. Soc. 2005, 127 (40), 13792−13793.(22) Son, J. S.; Choi, M. K.; Han, M. K.; Park, K.; Kim, J. Y.; Lim, S.J.; Oh, M.; Kuk, Y.; Park, C.; Kim, S. J.; Hyeon, T. Nano Lett. 2012, 12(2), 640−647.(23) Zebarjadi, M.; Shakouri, A.; Esfarjani, K. Phys. Rev. B 2006, 74(19), 195331.(24) Ko, D.-K.; Kang, Y.; Murray, C. B. Nano Lett. 2011, 11 (7),2841−2844.(25) Moyzhes, B.; Nemchinsky, V. Appl. Phys. Lett. 1998, 73 (13),1895−1897.(26) Mehta, R. J.; Zhang, Y.; Karthik, C.; Singh, B.; Siegel, R. W.;Borca-Tasciuc, T.; Ramanath, G. Nat. Mater. 2012, 11 (3), 233−240.(27) Bilc, D.; Mahanti, S. D.; Quarez, E.; Hsu, K.-F.; Pcionek, R.;Kanatzidis, M. G. Phys. Rev. Lett. 2004, 93 (14), 146403.(28) Kanatzidis, M. G. Chem. Mater. 2010, 22 (3), 648−659.(29) Yan, X.; Poudel, B.; Ma, Y.; Liu, W. S.; Joshi, G.; Wang, H.; Lan,Y.; Wang, D.; Chen, G.; Ren, Z. F. Nano Lett. 2010, 10 (9), 3373−3378.(30) Zhang, J.; Peng, Z.; Soni, A.; Zhao, Y.; Xiong, Y.; Peng, B.;Wang, J.; Dresselhaus, M. S.; Xiong, Q. Nano Lett. 2011, 11 (6),2407−2414.(31) Richter, W.; Becker, C. R. Physica Status Solidi B 1977, 84 (2),619−628.(32) Chen, Y. Y.; Yao, Y. D.; Lin, B. T.; Suo, C. T.; Shyu, S. G.; Lin,H. M. Nanostruct. Mater. 1995, 6 (5−8), 597−600.(33) Zhang, H. F.; Wu, X. J. Acta Phys. Sin. 1993, 2 (8), 583.(34) Souza, S. M.; Triches, D. M.; Poffo, C. M.; Lima, J. C. d.;Grandi, T. A.; Biasi, R. S. d. J. Appl. Phys. 2011, 109 (1), 013512.

Figure 6. Cross section TEM images showing weak links along thetwo directions: (a) top−bottom stackings and (b) head-to-headjoining. Scale bars are 5 nm.

Nano Letters Letter

dx.doi.org/10.1021/nl302017w | Nano Lett. 2012, 12, 4305−43104310